Future Cars: Electric and Autonomous Vehicles

[moda253]

What's not going to happen?

Notice I didn't say SOVs were going away, just the idea that every "normal" adult has to own one will likely change. Families owning "at least one vehicle" is not inconsistent with my argument.

I think that's the future, sharing cars and using other modes, rather than having every single adult purchase, insure, maintain, and store their own individual $10,000-$50,000 machine that seats up to five people just for commuting and errands.

Sorry it was when you said that you thought that the future would be more ride sharing and other modes rather than having every single person purchase, maintain,.......

I guess I just don't see any of that having a major dent in our transportation systems any time soon for many reasons. Would I support such a shift? Absolutely. Too many conveniences for too many people and the ability for automakers and others to make money off those conveniences tells me that isn't happening.

[Anondson]

Ford and Baidu are investing in a LIDAR company for car sensors.

https://techcrunch.com/2016/08/16/ford- ... ving-cars/

[mulad]

That reminds me that there has been some progress with solid-state LIDAR sensors. Currently-available sensors use laser beams bouncing off spinning mirrors, which can be bulky and are at risk of damage whenever the cars they're on go through potholes or other bumps. MIT and DARPA have gotten a prototype working -- It only has a range of about 2 meters so far, but they think they can improve it. A couple private companies are working on them too.

http://spectrum.ieee.org/tech-talk/semi ... -on-a-chip
https://techcrunch.com/2016/08/08/innov ... ate-lidar/

[moda253]

Huh... yeah we are decades away.....

http://www.usnews.com/news/business/art ... man-driver


DETROIT (AP) — Companies can now test self-driving cars on Michigan public roads without a driver or steering wheel under new laws that could push the state to the forefront of autonomous vehicle development.

The package of bills signed into law Friday comes with few specific state regulations and leaves many decisions up to automakers and companies like Google and Uber.

It also allows automakers and tech companies to run autonomous taxi services and permits test parades of self-driving tractor-trailers as long as humans are in each truck. And they allow the sale of self-driving vehicles to the public once they are tested and certified, according to the state.

The bills allow testing without burdensome regulations so the industry can move forward with potential life-saving technology, said Gov. Rick Snyder, who was to sign the bills. "It makes Michigan a place where particularly for the auto industry it's a good place to do work," he said.

The bills give Michigan the potential to be a leader by giving the companies more autonomy than say, California, which now requires human backup drivers in case something goes awry.

[Silophant]

Yep. Probably still decades away. Be careful walking in Michigan for a while though.

[nBode]

^Haven't autonomous vehicles been shown to be safer than human drivers?

[VAStationDude]

libertarian deregulation of a state's road network at the behest of said state's most prominent industry != technological break through. I hold fully autonomous personal vehicles will remain the dream of techno geeks throughout the land for the next several decades.

[mplsjaromir]

Governor Rick Snyder has an abysmal record of maintaining the safety of Michigan's citizens. Driverless cars are about as safe as Flint's drinking water.

[Silophant]

^Haven't autonomous vehicles been shown to be safer than human drivers?

So far, but a Mercedes exec recently said in an interview that a self-driving Mercedes will always prioritize the occupant's safety over a pedestrian's.

[Qhaberl]

That's scary, especially thinking about people who are blind or have disabilities. I'm not a huge fan of a computer determining who's life is more valuable, the driver or a pedestrian.


Sent from my iPhone using Tapatalk

[nBode]

^is that scarier than a human determining who's life is more valuable (in which case it would always be the driver of the vehicle, regardless of any other facts or circumstances)?

Maybe that's too philosophical.

[David Greene]

^is that scarier than a human determining who's life is more valuable (in which case it would always be the driver of the vehicle, regardless of any other facts or circumstances)?

I very much dispute that. Normal people really don't want to kill other people. The scary part is how does the computer decide that it's a lose-lose situation? It's got to be exceedingly rare. I very much trust a human to assess the situation more accurately than a computer. I work on computers daily.

[nBode]

I work with humans daily. We're very flawed.
(sort've joking, sort've not)

Computers don't ever want to kill people (at least, that's how we will program them). The computer decides based off of the way we design them. If we design them to always prioritize the driver (which is how humans operate motor vehicles, and probably the most likely outcome), they will do so. If we can design a system which overall performs better than our current, human-operated system (which I believe we can), then there is nothing more-scary about it.

I don't think it likely that automated cars will eliminate traffic fatalities, and there will certainly arise situations which prompt reconsideration of ever-changing laws and procedures. Theoretically, a computer would be more apt to being objective in fatal situations than a human would (This delves into the ethics of the value of individual human life, an entire topic of its own).

In any event, if we create a system which analyses data, then reacts to that data in the manner which, to its knowledge, is most likely to preserve occupant life (i.e. how humans drive) AND such a system performs well enough to substantially reduce the likelihood of such scenarios, then there is less to be scared about than there is now.

I can understand being afraid of things we don't understand (that is normal), such as a computer's reasoning behind its decisions. But be honest, do you really understand humans' reasonings behind our decisions?

[MNdible]

Setting aside all of the above, I'd just note that a self-driving car system will not be distracted, will not fall asleep, and will not be drunk. And even if you assume that a human driver is in no way impaired, the self-driving car will still have much faster reaction time.

[David Greene]

I'm not worried about that sort of stuff. I'm worried about how the computer would assess a specific situation and decide someone must die.

If someone dies, who's legally responsible?

[kirby96]

I'm not worried about that sort of stuff. I'm worried about how the computer would assess a specific situation and decide someone must die.

If someone dies, who's legally responsible?

That would be a real problem. It would also be a real problem that hardly ever occurs. I suspect it hardly ever occurs now.

...not saying it never occurs, but I gotta think the vast majority (like probably 99% or more) of innocent bystander injuries and deaths are due to causes things automated cars will improve (drunk, inattentive, sleeping) as opposed to an otherwise law-abiding attentive driver (which a computer would be) reacting to an unforeseen circumstance in order to save ones self.

Basically, this seems like a risk we can deal with. I mean, the risk of getting hurt by an uninsured driver is very real today, and if you sue there's a good chance you will recover nothing.

[David Greene]

Mostly true, but I can guarantee you there *will* be bugs that cause the car to do something bad a human would not do. I'm not saying scrap the whole thing but rather temper the optimism a bit.

[EOst]

Even for "innocent bystander" incidents (i.e. without the pedestrian or cyclist doing anything dangerous, stupid, or illegal) 99% seems very high to me. If we had a dollar for every pedestrian struck because a driver swerved to avoid hitting a cat, we'd be rich as kings.

The human impulse to assign blame is a powerful thing, one that shouldn't be easily dismissed. We know from Demosthenes that in classical Athens, if a stone fell from a wall and killed someone, that stone would be publicly tried for murder.

If and when a self-driving car strikes and kills a pedestrian, the calls for new regulations will be almost overwhelming. When an airplane crashes, aeronautics companies make no headway by pointing out that, statistically, flying is still safer than driving. Self-driving cars will have no better luck.

[kirby96]

Mostly true, but I can guarantee you there *will* be bugs that cause the car to do something bad a human would not do. I'm not saying scrap the whole thing but rather temper the optimism a bit.

No argument on that. To your point, clearly the underlying assumption here is that self-driving cars won't be deployed unless they are definitively safer than human counterparts (like, at least an order of magnitude safer and probably multiple orders of magnitude). Maybe the real question is how do we actually come to that conclusion?

...and to my point about the risk, presuming that in aggregate the self driving cars are that much safer overall, I could see it being not that big a deal for manufacturers to insure against that risk in the same manner they do whatever they do regarding general product liability. Basically, if it's a rare enough occurrence, it's probably going to be far less costly than, say, liability for faulty air-bags, unintended acceleration, exploding Pintos, or any of the other product faults that have arisen over the years.

[kirby96]

Even for "innocent bystander" incidents (i.e. without the pedestrian or cyclist doing anything dangerous, stupid, or illegal) 99% seems very high to me. If we had a dollar for every pedestrian struck because a driver swerved to avoid hitting a cat, we'd be rich as kings.

Well, I admit I don't know the actual stats, but I really don't think it actually happens that often (whether it's 80% or 99%). When I think of the pedestrian incidents I've heard in Minneapolis over the past couple years, it's almost always inattention, drunk, or someone claiming someone else (driver, pedestrian, cyclist) done screwed up.

But most importantly, that's precisely the situation a computer will avoid.

The human impulse to assign blame is a powerful thing, one that shouldn't be easily dismissed. We know from Demosthenes that in classical Athens, if a stone fell from a wall and killed someone, that stone would be publicly tried for murder.

If and when a self-driving car strikes and kills a pedestrian, the calls for new regulations will be almost overwhelming. When an airplane crashes, aeronautics companies make no headway by pointing out that, statistically, flying is still safer than driving. Self-driving cars will have no better luck.

But the aeronautics companies HAVE won that battle. Go down to MSP and watch planes take off every 30 seconds or so. When planes crash we don't see calls for regulations to make it 100% safe, we have a system of review and determination of fixes, and then it goes on. The Dreamliner had a very bumpy launch and they fly everyday. The airlines/aeronautics companies simply absorb the costs of the very (very) limited risk.

I acknowledge there is a psychological hurdle here. It will seem 'scary', but so did a lot of tech that is now ubiquitous.